Before children will become productive with language they have already learned something about word order and grammatical rules. Children of 16 months of age are already sensitive to word order ( Akhtar, Tomasello, 1997). This is an outcome of the preferential looking paradigma. In the study of Hirsch-Paek and Goliknhoff (1996), children were shown two films at the same time. The differences between these films were; that in one film Big Bird was washing Cookie Monster, in the other Cookie Monster was washing Big Bird. When the infants were asked 'where is Big Bird washing Cookie Monster?', the infants spend more time looking at the film representing this action (Siegler, Wagner Alibali, 2005).
As discussed before infants do possess some information about grammatical rules, however, they will only become productive with language at the age of two years. Productivity is defined as the ability to generalize linguistic knowledge ( Akhtar, Tomasello, 1997; Tomasello, Akhtar, Dodson, Rekau, 1997). Productivity with language means that the child knows something about abstract linguistic structures in the form of categories, schemas, analogies or rules ( Tomasello, Akhtar, Dodson, Rekau, 1997) Dutch examples of morphological productivity include adding a plural -s, -es to the end of a noun that has been heard only in the singular form. Like telefoon- telefoons (telephone- telephones). Or adding -en as a way to express a plural, boek- boeken (book-books). In English morphological productivity includes adding a plural - s, e.g. cat- cats. Productivity with language is also about verbal inflections. In Dutch examples of morphological productivity with inflections includes adding a -t or -en to the end of a verb that has been heard only in the bare stem, so the past participle is formed. In English the past participle is formed by adding -ing, ed tot the bare stem, walk- walked ( Akhtar, Tomasello, 1997).
Children learn morphological and inflections trough modelling. However, children make mistakes, for instance, over regularization errors. These errors are one source of evidence proving that children clearly use their language productively in a number of ways that indicate a grammatical category of a verb (Olguin, Tomasello, 1993). This also shows the productivity in plurals. Overregularization errors can be seen as speech errors in which children treat abnormal forms of words as if they were regular ( Siegler, Deloache, Eisenberg, 2006; Siegler, Alibali, 2005). For instance, children say tafelen (tables) instead of the correct plural tafels. They also do this with verb inflection. Children treat the irregular form of inflection as if they were regular. For instance, children say geloopt (walked) instead of the correct inflection of lopen, gelopen. Overrularization errors are interesting because parents almost never overregularize. It shows that language learning involves more that only imitation. Children must own mechanisms that detect and extend linguistic generalizations (Marcus, 1996). Toddlers need to learn much about their language, nevertheless, they represent language experience in terms of an abstract mental vocabulary. This abstract mental vocabulary makes it possible for children to detect general patterns in their language quickly, therefore, this makes it possible tot learn rules and words (Gertner, Fisher, Eisengart, 2006). When children are around two years of age, they are able to make the distinction between nouns and verbs ( Hohenstein & Akhtar, 2007; Tomasello, Akhtaer, Dodson and Rekau, 1997). They don't drop -ing from nouns and understand that it is inappropriate to do so. Additionally, they do sometimes drop -ing from verbs and know that it is appropriate to do so ( Hohenstein & Akhtar, 2007).
Verbs play a important role in early grammatical development (Olguin, Tomasselo, 1993). According to Tomasello (1992) and Tomasello & Brooks (1999) young children do not produce transitive utterances on the basis of abstract and adult like constructions. However, children do produce transitive or any other kind of utterances on the basis of an inventory of item specific schemas, which all are characterized by the specific predicate involved (Tomasello, 1992). Tomasello calls this theory the 'verb island hypothesis'. According to the 'verb island hypothesis' children develop a separate minisyntaxis for each verb with its own semantic rules. This minisyntaxis is independent of other verbs, it is verb-specific, children organize their grammar rules around separate verbs. Therefore, they do not organize their grammar around more general sentence-schemes such as subject-verb-object scheme (Olguin,Tomasello, 1993). In the sentence Marieke doet Kimberley pijn (Marieke is hurting Kimberley) , Marieke and Kimberley are not classified as subject and object. They are seen as someone who is hurting someone else and as someone who can be hurt. Different structures are build around particular verbs. Simple patterns will be learned by imitation and children will only use the verb inflection after they have heard it. They produce a single unmarked form for every verb (Marchman, Bates, 1992). When children start producing their first past tense forms, they produce the correct morphological mappings associated with individual regular verbs and irregular verbs. For example in dutch the regular verb werken (work), werk- werkte, and the irregular verb slapen (sleep), slaap - sliep will both be used correctly. However, children do this without extracting a general pattern. (Marchman, Bates, 1992). That will happen once they have gained enough experience with other verb inflections. Children will build up these experiences slowly as they hear different verb inflections at the same time. Generalization will take place at a later stadium. Children build up constructions by collecting some 'critical mass' of transitive verb island constructions (Childers, Tomasello, 2001). This 'critical mass' is the basis for the aforementioned generalizations. Children gradually expend this verb by verb to make it applicable across a wider range of verbs (Roberts, 1983), creating an interconnected verbal system.
However, Nino (2003) does not agree with this 'verb island hypotheses. Nino (2003) sees the developing of children's early grammar not as an collection of different verb-islands, but as a system that is present from day one (Nino, 2003). She does not support the insularity of early syntactic development that was claimed by Tomasello (YEAR). The view of Nino (2003) is that there are considerable transfers and facilitations between verbs. Children do generalize their knowledge to other verbs. They do not build up complex structures on a completely item-specific manner. Novel verb constructions do not need to have prior and less complex constructions in the child's speech (Nino,2003). It is possible that novel verb construction has a related prior and less complex construction; however, it is not necessary for the course of development. Where in tomasello's (YEAR) view the development of children's grammar can be seen as a collection of verb-islands, a web with all items connected to each other, be a metaphor for the view of Nino (2003). In this view, all the verbs are interconnected and children do generalize their knowledge to other verbs.
Akhtar and Tomasello's (1997) research focused on young children's productivity with word order and verbal morphology. The findings of this research will be discussed in terms of Tomasello's verb island hypothesis. The -ing inflection was used productively before the ÂÂ-ed inflections. -Ing was used productively when children were three years old and -ed was used productively when children were at the end of their third year. Older children also comprehend the word order with the novel words, whereas younger children dit not. The studies also suggest that children younger than three years of age have most probably learned the appropriate use of word order on a verb-by verb basis. This all is supporting the verb island hypothesis of Tomasello (YEAR).
Childres and Tomasello (2001) did research focusing on the role of pronouns in the acquisition of the English transitive construction by young children. Their research does not support the verb island hypothesis. Fifty children of two years and 6 months of age ( M=2 years 6 months two weeks; range two years 4 months to two years and 10 months) participated in the study. In this study children played sixteen different games. Each game was associated with its own verb and involved a highly transitive action that could be named by either of two English transitive verbs. This means that an agent initiates an action that changes the state of the patient. One of these verbs was likely to be known by the children, while the other verb was expected to be unfamiliar. Each of the fifty participating children were randomly assigned to one of four training conditions. The conditions where defined by the crossing of verb familiarity (familiar, unfamiliar) with the manner in which agent and patient were labelled (nouns, pronouns). The fifth group was a control group. Thus the five conditions of the experiment were as follows: (1) familiar verb + nouns (e.g. 'Look! The bird is swinging the bathtub. See? The bird is swinging the bathtub'), (2) Unfamiliar verb + noun (e.g. Look! the dog is hurling the chair. See? The dog is hurling the chair', (3) familiar verb + pronouns (e.g. Look! the cow is pulling the car. See? He is pulling it?) , (4) Unfamiliar verb + pronouns (e.g. Look! the bear is striking the tree. See? He is striking it, (5) control group. All the children participated in three training sessions and then one testing setting. Each session took place on a separate day and lasted twenty to twenty-five minutes. The children were asked tot select a toy character to play the game with. The experimenter then acted out the action with the toy character and described it in the right manner for the experimental condition (see above). Each event was acted out six times before the children moved to the next game. The children heard a total of twelve uses of each verb on each day. When the first three actions where acted out by the experimenter, the child was invited to use the toy character to act out the event in order to maintain interest, and also to see whether the child uses the verb in its spontaneous speech. The experimenter asked the child after the final three actions where acted out; 'What is happening' and 'What is (agent) doing'. Herewith, the experimenter tries to elict production of the verb. Results of this study show that during training, children mainly produced the familiar verbs and show an affection to the most often produced verbs trained with pronouns. Children who heard pronouns during training produces almost twice as many transitive utterances in the test trials as the children who only heard nouns. There was no effect found for verb type. The verb island hypothesis was not supported in this study. Strong evidence was found for the meaningful role of pronouns.